--- name: "ssh_sftp" section: "3erl" platform: "linux" package: "erlang-manpages" version: "1:25.2.3+dfsg-1+deb12u4" release: "Debian 12 (Bookworm)" platforms: ["linux"] url: "https://new.carta.tech/ssh_sftp" --- # ssh\_sftp(3erl) SFTP client. ## NAME ssh\_sftp - SFTP client. ## DESCRIPTION This module implements an SSH FTP (SFTP) client. SFTP is a secure, encrypted file transfer service available for SSH. ## DATA TYPES ``` sftp_option() = {timeout, timeout()} | {sftp_vsn, integer() >= 1} | {window_size, integer() >= 1} | {packet_size, integer() >= 1} ``` ### Error cause ``` reason() = atom() | string() | tuple() ``` A description of the reason why an operation failed. The *atom()* value is formed from the sftp error codes in the protocol-level responses as defined in draft-ietf-secsh-filexfer-13 section 9.1. The codes are named as *SSH\_FX\_\** which are transformed into lowercase of the star-part. E.g. the error code *SSH\_FX\_NO\_SUCH\_FILE* will cause the *reason()* to be *no\_such\_file*. The *string()* reason is the error information from the server in case of an exit-signal. If that information is empty, the reason is the exit signal name. The *tuple()* reason are other errors like for example *{exit\_status,1}*. ### Crypto operations for open\_tar ``` tar_crypto_spec() = encrypt_spec() | decrypt_spec() ``` ``` encrypt_spec() = {init_fun(), crypto_fun(), final_fun()} ``` ``` decrypt_spec() = {init_fun(), crypto_fun()} ``` Specifies the encryption or decryption applied to tar files when using open\_tar/3 or open\_tar/4. The encryption or decryption is applied to the generated stream of bytes prior to sending the resulting stream to the SFTP server. For code examples see Section Example with encryption in the ssh Users Guide. ``` init_fun() = fun(() -> {ok, crypto_state()}) | fun(() -> {ok, crypto_state(), chunk_size()}) ``` ``` chunk_size() = undefined | integer() >= 1 ``` ``` crypto_state() = any() ``` The *init\_fun()* in the tar\_crypto\_spec is applied once prior to any other *crypto* operation. The intention is that this function initiates the encryption or decryption for example by calling crypto:crypto\_init/4 or similar. The *crypto\_state()* is the state such a function may return. If the selected cipher needs to have the input data partitioned into blocks of a certain size, the *init\_fun()* should return the second form of return value with the *chunk\_size()* set to the block size. If the *chunk\_size()* is *undefined*, the size of the *PlainBin*s varies, because this is intended for stream crypto, whereas a fixed *chunk\_size()* is intended for block crypto. A *chunk\_size()* can be changed in the return from the *crypto\_fun()*. The value can be changed between *pos\_integer()* and *undefined*. ``` crypto_fun() = fun((TextIn :: binary(), crypto_state()) -> crypto_result()) ``` ``` crypto_result() = {ok, TextOut :: binary(), crypto_state()} | {ok, TextOut :: binary(), crypto_state(), chunk_size()} ``` The initial *crypto\_state()* returned from the init\_fun() is folded into repeated applications of the *crypto\_fun()* in the tar\_crypto\_spec. The binary returned from that fun is sent to the remote SFTP server and the new *crypto\_state()* is used in the next call of the *crypto\_fun()*. If the *crypto\_fun()* reurns a *chunk\_size()*, that value is as block size for further blocks in calls to *crypto\_fun()*. ``` final_fun() = fun((FinalTextIn :: binary(), crypto_state()) -> {ok, FinalTextOut :: binary()}) ``` If doing encryption, the *final\_fun()* in the tar\_crypto\_spec is applied to the last piece of data. The *final\_fun()* is responsible for padding (if needed) and encryption of that last piece. ## EXPORTS ``` apread(ChannelPid, Handle, Position, Len) -> {async, N} | Error ``` Types: ChannelPid = pid()\ Handle = term()\ Position = Len = integer()\ Error = {error, reason()}\ N = term()\ The *apread/4* function reads from a specified position, combining the *position/3* and *aread/3* functions. ``` apwrite(ChannelPid, Handle, Position, Data) -> {async, N} | Error ``` Types: ChannelPid = pid()\ Handle = term()\ Position = integer()\ Data = binary()\ Error = {error, reason()}\ N = term()\ The *apwrite/4* function writes to a specified position, combining the *position/3* and *awrite/3* functions. ``` aread(ChannelPid, Handle, Len) -> {async, N} | Error ``` Types: ChannelPid = pid()\ Handle = term()\ Len = integer()\ Error = {error, reason()}\ N = term()\ Reads from an open file, without waiting for the result. If the handle is valid, the function returns *{async, N}*, where *N* is a term guaranteed to be unique between calls of *aread*. The actual data is sent as a message to the calling process. This message has the form *{async\_reply, N, Result}*, where *Result* is the result from the read, either *{ok, Data}*, *eof*, or *{error, reason()}*. ``` awrite(ChannelPid, Handle, Data) -> {async, N} | Error ``` Types: ChannelPid = pid()\ Handle = term()\ Data = binary()\ Error = {error, reason()}\ N = term()\ Writes to an open file, without waiting for the result. If the handle is valid, the function returns *{async, N}*, where *N* is a term guaranteed to be unique between calls of *awrite*. The result of the *write* operation is sent as a message to the calling process. This message has the form *{async\_reply, N, Result}*, where *Result* is the result from the write, either *ok*, or *{error, reason()}*. ``` close(ChannelPid, Handle) -> ok | Error ``` ``` close(ChannelPid, Handle, Timeout) -> ok | Error ``` Types: ChannelPid = pid()\ Handle = term()\ Timeout = timeout()\ Error = {error, reason()}\ Closes a handle to an open file or directory on the server. ``` delete(ChannelPid, Name) -> ok | Error ``` ``` delete(ChannelPid, Name, Timeout) -> ok | Error ``` Types: ChannelPid = pid()\ Name = string()\ Timeout = timeout()\ Error = {error, reason()}\ Deletes the file specified by *Name*. ``` del_dir(ChannelPid, Name) -> ok | Error ``` ``` del_dir(ChannelPid, Name, Timeout) -> ok | Error ``` Types: ChannelPid = pid()\ Name = string()\ Timeout = timeout()\ Error = {error, reason()}\ Deletes a directory specified by *Name*. The directory must be empty before it can be successfully deleted. ``` list_dir(ChannelPid, Path) -> {ok, FileNames} | Error ``` ``` list_dir(ChannelPid, Path, Timeout) -> {ok, FileNames} | Error ``` Types: ChannelPid = pid()\ Path = string()\ Timeout = timeout()\ FileNames = \[FileName\]\ FileName = string()\ Error = {error, reason()}\ Lists the given directory on the server, returning the filenames as a list of strings. ``` make_dir(ChannelPid, Name) -> ok | Error ``` ``` make_dir(ChannelPid, Name, Timeout) -> ok | Error ``` Types: ChannelPid = pid()\ Name = string()\ Timeout = timeout()\ Error = {error, reason()}\ Creates a directory specified by *Name*. *Name* must be a full path to a new directory. The directory can only be created in an existing directory. ``` make_symlink(ChannelPid, Name, Target) -> ok | Error ``` ``` make_symlink(ChannelPid, Name, Target, Timeout) -> ok | Error ``` Types: ChannelPid = pid()\ Name = Target = string()\ Timeout = timeout()\ Error = {error, reason()}\ Creates a symbolic link pointing to *Target* with the name *Name*. ``` open(ChannelPid, Name, Mode) -> {ok, Handle} | Error ``` ``` open(ChannelPid, Name, Mode, Timeout) -> {ok, Handle} | Error ``` Types: ChannelPid = pid()\ Name = string()\ Mode = \[read \| write \| append \| binary \| raw\]\ Timeout = timeout()\ Handle = term()\ Error = {error, reason()}\ Opens a file on the server and returns a handle, which can be used for reading or writing. ``` opendir(ChannelPid, Path) -> {ok, Handle} | Error ``` ``` opendir(ChannelPid, Path, Timeout) -> {ok, Handle} | Error ``` Types: ChannelPid = pid()\ Path = string()\ Timeout = timeout()\ Handle = term()\ Error = {error, reason()}\ Opens a handle to a directory on the server. The handle can be used for reading directory contents. ``` open_tar(ChannelPid, Path, Mode) -> {ok, Handle} | Error ``` ``` open_tar(ChannelPid, Path, Mode, Timeout) -> {ok, Handle} | Error ``` Types: ChannelPid = pid()\ Path = string()\ Mode = \[read \| write \| {crypto, tar\_crypto\_spec()}\]\ Timeout = timeout()\ Handle = term()\ Error = {error, reason()}\ Opens a handle to a tar file on the server, associated with *ChannelPid*. The handle can be used for remote tar creation and extraction. The actual writing and reading is performed by calls to erl\_tar:add/3,4 and erl\_tar:extract/2. Note: The erl\_tar:init/3 function should not be called, that one is called by this open\_tar function. For code examples see Section SFTP Client with TAR Compression in the ssh Users Guide. The *crypto* mode option is explained in the data types section above, see Crypto operations for open\_tar. Encryption is assumed if the *Mode* contains *write*, and decryption if the *Mode* contains *read*. ``` position(ChannelPid, Handle, Location) -> {ok, NewPosition} | Error ``` ``` position(ChannelPid, Handle, Location, Timeout) -> {ok, NewPosition} | Error ``` Types: ChannelPid = pid()\ Handle = term()\ Location = \ \ Offset \|\ \ {bof, Offset} \|\ \ {cur, Offset} \|\ \ {eof, Offset} \|\ \ bof \| cur \| eof\ Timeout = timeout()\ Offset = NewPosition = integer()\ Error = {error, reason()}\ Sets the file position of the file referenced by *Handle*. Returns *{ok, NewPosition}* (as an absolute offset) if successful, otherwise *{error, reason()}*. *Location* is one of the following: - ***Offset*:** The same as *{bof, Offset}*. - ***{bof, Offset}*:** Absolute offset. - ***{cur, Offset}*:** Offset from the current position. - ***{eof, Offset}*:** Offset from the end of file. - ***bof \| cur \| eof*:** The same as eariler with *Offset* 0, that is, *{bof, 0} \| {cur, 0} \| {eof, 0}*. ``` pread(ChannelPid, Handle, Position, Len) -> {ok, Data} | eof | Error ``` ``` pread(ChannelPid, Handle, Position, Len, Timeout) -> {ok, Data} | eof | Error ``` Types: ChannelPid = pid()\ Handle = term()\ Position = Len = integer()\ Timeout = timeout()\ Data = string() \| binary()\ Error = {error, reason()}\ The *pread/3,4* function reads from a specified position, combining the *position/3* and *read/3,4* functions. ``` pwrite(ChannelPid, Handle, Position, Data) -> ok | Error ``` ``` pwrite(ChannelPid, Handle, Position, Data, Timeout) -> ok | Error ``` Types: ChannelPid = pid()\ Handle = term()\ Position = integer()\ Data = iolist()\ Timeout = timeout()\ Error = {error, reason()}\ The *pwrite/3,4* function writes to a specified position, combining the *position/3* and *write/3,4* functions. ``` read(ChannelPid, Handle, Len) -> {ok, Data} | eof | Error ``` ``` read(ChannelPid, Handle, Len, Timeout) -> {ok, Data} | eof | Error ``` Types: ChannelPid = pid()\ Handle = term()\ Len = integer()\ Timeout = timeout()\ Data = string() \| binary()\ Error = {error, reason()}\ Reads *Len* bytes from the file referenced by *Handle*. Returns *{ok, Data}*, *eof*, or *{error, reason()}*. If the file is opened with *binary*, *Data* is a binary, otherwise it is a string. If the file is read past *eof*, only the remaining bytes are read and returned. If no bytes are read, *eof* is returned. ``` read_file(ChannelPid, File) -> {ok, Data} | Error ``` ``` read_file(ChannelPid, File, Timeout) -> {ok, Data} | Error ``` Types: ChannelPid = pid()\ File = string()\ Data = binary()\ Timeout = timeout()\ Error = {error, reason()}\ Reads a file from the server, and returns the data in a binary. ``` read_file_info(ChannelPid, Name) -> {ok, FileInfo} | Error ``` ``` read_file_info(ChannelPid, Name, Timeout) -> {ok, FileInfo} | Error ``` Types: ChannelPid = pid()\ Name = string()\ Timeout = timeout()\ FileInfo = file:file\_info()\ Error = {error, reason()}\ Returns a *file\_info* record from the file system object specified by *Name* or *Handle*. See file:read\_file\_info/2 for information about the record. Depending on the underlying OS:es links might be followed and info on the final file, directory etc is returned. See read\_link\_info/2 on how to get information on links instead. ``` read_link(ChannelPid, Name) -> {ok, Target} | Error ``` ``` read_link(ChannelPid, Name, Timeout) -> {ok, Target} | Error ``` Types: ChannelPid = pid()\ Name = Target = string()\ Timeout = timeout()\ Error = {error, reason()}\ Reads the link target from the symbolic link specified by *name*. ``` read_link_info(ChannelPid, Name) -> {ok, FileInfo} | Error ``` ``` read_link_info(ChannelPid, Name, Timeout) -> {ok, FileInfo} | Error ``` Types: ChannelPid = pid()\ Name = string()\ FileInfo = file:file\_info()\ Timeout = timeout()\ Error = {error, reason()}\ Returns a *file\_info* record from the symbolic link specified by *Name* or *Handle*. See file:read\_link\_info/2 for information about the record. ``` rename(ChannelPid, OldName, NewName) -> ok | Error ``` ``` rename(ChannelPid, OldName, NewName, Timeout) -> ok | Error ``` Types: ChannelPid = pid()\ OldName = NewName = string()\ Timeout = timeout()\ Error = {error, reason()}\ Renames a file named *OldName* and gives it the name *NewName*. **start\_channel(ConnectionRef) -\>**\ **start\_channel(ConnectionRef, SftpOptions) -\> {ok, ChannelPid} \| Error**\ **start\_channel(Host) -\>**\ **start\_channel(Host, Options) -\>**\ **start\_channel(Host, Port, Options) -\>**\ **start\_channel(TcpSocket) -\>**\ **start\_channel(TcpSocket, Options) -\> {ok, ChannelPid, ConnectionRef} \| Error**\ Types: Host = ssh:host()\ Port = inet:port\_number()\ TcpSocket = ssh:open\_socket()\ Options = \[ sftp\_option() \| ssh:client\_option() \]\ SftpOptions = \[ sftp\_option() \]\ ChannelPid = pid()\ ConnectionRef = ssh:connection\_ref()\ Error = {error, reason()}\ If no connection reference is provided, a connection is set up, and the new connection is returned. An SSH channel process is started to handle the communication with the SFTP server. The returned *pid* for this process is to be used as input to all other API functions in this module. Options: - ***{timeout, timeout()}*:** There are two ways to set a timeout for the underlying ssh connection: - \* If the connection timeout option *connect\_timeout* is set, that value is used also for the negotiation timeout and this option (*timeout*) is ignored. - \* Otherwise, this option (*timeout*) is used as the negotiation timeout only and there is no connection timeout set The value defaults to *infinity*. - ***{sftp\_vsn, integer()}*:** Desired SFTP protocol version. The actual version is the minimum of the desired version and the maximum supported versions by the SFTP server. All other options are directly passed to ssh:connect/3 or ignored if a connection is already provided. ``` stop_channel(ChannelPid) -> ok ``` Types: ChannelPid = pid()\ Stops an SFTP channel. Does not close the SSH connection. Use ssh:close/1 to close it. ``` write(ChannelPid, Handle, Data) -> ok | Error ``` ``` write(ChannelPid, Handle, Data, Timeout) -> ok | Error ``` Types: ChannelPid = pid()\ Handle = term()\ Data = iodata()\ Timeout = timeout()\ Error = {error, reason()}\ Writes *data* to the file referenced by *Handle*. The file is to be opened with *write* or *append* flag. Returns *ok* if successful or *{error, reason()}* otherwise. ``` write_file(ChannelPid, File, Data) -> ok | Error ``` ``` write_file(ChannelPid, File, Data, Timeout) -> ok | Error ``` Types: ChannelPid = pid()\ File = string()\ Data = iodata()\ Timeout = timeout()\ Error = {error, reason()}\ Writes a file to the server. The file is created if it does not exist but overwritten if it exists. ``` write_file_info(ChannelPid, Name, FileInfo) -> ok | Error ``` ``` write_file_info(ChannelPid, Name, FileInfo, Timeout) -> ok | Error ``` Types: ChannelPid = pid()\ Name = string()\ FileInfo = file:file\_info()\ Timeout = timeout()\ Error = {error, reason()}\ Writes file information from a *file\_info* record to the file specified by *Name*. See file:write\_file\_info/\[2,3\] for information about the record.